Microsoft and OpenAI are allegedly developing a 100 billion dollar data centre project. The Information published a report on Friday, March 29th, stating that the centre will house an AI supercomputer known as “Stargate,” which is anticipated to go live by 2028.
The article quotes insiders involved in the initiative’s debates. It states that Microsoft would finance the project, which is anticipated to be far more expensive than even the most significant data centres. The two technology firms intend to construct several supercomputers, the largest of which would be the Stargate.
According to the study, OpenAI’s capacity to release the upcoming big update, which is anticipated to be released early next year, would significantly impact Stargate’s progress.
PYMNTS has requested feedback from both companies but has yet to respond. The “fight for generative AI” erupted after OpenAI unveiled its ChatGPT model, which was reviewed by PYMNTS last week. John Licato, a lecturer in the engineering and computer science department at the University of South Florida, informed PYMNTS that Google is one of the potential rivals for this position.
He explained that larger contexts may be achieved using Google’s Gemini models, especially Gemini 1.5 Pro, which comes with an operational window worth approximately a million tokens, in contrast to GPT-4’s 128,000 token limitation. He related this to the company’s “institutional experience” and “access to computational resources and information.”
Licato went on to say that Google has access to data that few organizations can match, in addition to having extensive experience with transformers, the technology at the core of ChatGPT. He mentioned Anthropic and Meta as other powerful rivals.
Licato stated that access to massive processing power may be the most crucial element. Businesses such as Google and OpenAI own GPU processors worth millions, if not billions of dollars, in addition to more sophisticated technology for computing as TPUs (tensor processing units).
PYMNTS looked at Openstream last week, among other things, in the artificial intelligence space. With the company’s enterprise virtual assistant (EVA) system, which generates artificial intelligence characters, automated assistants, and voice drivers that can converse like humans without the need for complicated back-end programming or the possibility of illusions, artificial intelligence (AI) has recently received a patent for an application designed to facilitate business-to-business communication.
Raj Tumuluri, the president and chief executive officer of Openstream, informed PYMNTS in an interview that our unstructured strategy for speech control stands apart from other suppliers because we are not dependent on a planned dialogue script. We obtain tremendously powerful capabilities by combining the capacity to analyze and strategize about these actions, circumstances, and things in dialogue (the symbolic part) with the neural component by employing LLMs for understanding these acts, circumstances, and objects.
